JP4468294B2 - EXPERIENCE INFORMATION EVALUATION DEVICE, PROGRAM, AND COMPUTER-READABLE RECORDING MEDIUM - Google Patents

Description

The present invention relates to experience information Review AtaiSo location and program and a computer-readable recording medium, in particular, from the circumstances, such as all kinds of information, such as the Internet are mixed, analyzes the sentence included in the information and can provide experiences related keywords of interest relates to experience information Review AtaiSo location and program and a computer-readable recording medium utilizing a target keyword.

  In recent years, many people have published their opinions on the Internet regarding various matters such as hotels, travel, and electrical appliances. This public place includes a bulletin board where opinions of various people gathered, such as a shopping site itself, a weblog where personal opinions are gathered, and a personal homepage. In particular, the user's voice based on actual experience is highly useful as support for marketing, product development, product purchase consideration, etc. as word-of-mouth information or reputation information.

  For this reason, there is a technique for determining whether or not the information describes that the user actually experienced a certain thing. In this method, the past form of a verb, which is a descriptive feature that appears when describing what has been experienced, is used as an experience expression, and the sentence in which it is described is determined as experience information.

  When using the above technique, morphological analysis of sentences is often performed. As a technique of this morpheme analysis, for example, there is a Japanese morpheme analysis system that reads homomorphic words using word co-occurrence. Since co-occurrence of words is easy to identify for a person, it is difficult to make a mistake and the accuracy of the system can be reliably increased (see, for example, Non-Patent Document 1).

  “Experience information” handled in this case refers to information describing that the user actually experienced a certain thing. An example is shown in FIG. In Fig. 12, A, C, and D are like "I'm doing", "I'm eating", "I've been" about what the writer actually experienced or continuously experiencing It expresses using simple experience expression. Also, B and E in FIG. 12 express opinions and impressions obtained as a result of the actual experience of the writer using experience expressions such as “good” and “bad”.

  Examples that are not experiential expressions include those shown in F and G of FIG.

  F in FIG. 12 indicates only the condition that the machine is cheap, and it cannot be determined whether the writer has experienced it. As for G, it is not possible to specify whether or not the writer has experienced only by showing the conditions of amenity and fee.

The experience information may be a long sentence, or may be expressed in one sentence as shown in A to E of FIG.
Takeshi Tsuji, et al., “Japanese Morphological Analysis System that Considers Conservativeness” Report of the Information Processing Society of Japan Natural Language Processing Study Group 117-09, vol. 1997-no, 004, pp59-66

  When the user wants to find the information he wants, he uses some keyword and searches using a database or a search engine where that information is stored. When a user searches for experience information about a certain thing, if a keyword search is performed as described above, a lot of information other than the experience information such as advertisements and explanation of the keyword will be mixed, and it will arrive at the desired information. Is difficult. In particular, in recent years, there are many weblogs and bulletin boards that contain personal opinions and so on, so there are a large number of documents with mixed talk topics. Even if the information source is limited to only experience information, experience information that has nothing to do with keywords is picked up, and it is still difficult to reach the information that the user wants.

For example,
Example 1) “Today I went to a restaurant. For the Thanksgiving Day, I was able to eat a 10,000 yen course for 5000 yen. It was delicious, cheap, and I was very satisfied. I want mobile phone XXX. "
If there is information like This can be said to be “restaurant” experience information because it includes an expression representing the experience “I went” for “restaurant”. However, even if the user searches for information on the “mobile phone”, the information is presented.

  Further, according to the conventional technology, it is possible to determine whether or not a certain text document is experience information, but it is not possible to determine whether or not it is experience information regarding a certain keyword.

The present invention has been made in view of the above, using the experience representation corresponding to the keyword that the user pays attention, experience information Review AtaiSo location capable of prioritizing document set associated with that keyword Another object of the present invention is to provide a program and a computer-readable recording medium.

  FIG. 1 is an explanatory diagram of the principle of the present invention.

This onset Ming, keyword of interest (below, say) an experience information evaluation method using the,
When using a phrase that expresses the use or experience / experience of things as an experience expression,
The experience expression selection processing means, when an attention keyword is inputted, refers to the experience expression dictionary storage means in which the relationship between the attention keyword and the experience expression is stored, selects the experience expression related to the attention keyword, Experience expression selection step (step 1) to be stored in the experience expression storage means;
The keyword distance measurement processing means determines the appearance position relationship between the keyword of interest and the experience expression stored in the experience expression storage means selected in the experience expression selection step in each text document of the inputted input text document group. Keyword distance measurement step to measure (Step 2),
The evaluation means evaluates each text document in the input text document group based on the relationship between the attention keyword and the experience expression and the appearance position relation between the experience expression measured by the keyword distance measurement step, and the evaluation result and the input text. An evaluation step (step 3) for associating the document and outputting it to the document storage means ;
In the evaluation step (step 3),
Based on the strength of the relationship between the keyword of interest and the experience expression, a ranking calculation step is performed to determine the order relationship between the text documents in the input text document group by the ranking evaluation formula,
In the ranking calculation step,
As a ranking evaluation formula,
The value changes depending on the location of the attention keyword and experience expression,
If the experience text related to the keyword of interest is described before the keyword of interest in the input text document, the value of the evaluation formula will be low,
In the input text document, when an experiential expression related to the keyword of interest is described after the keyword of interest, an evaluation expression that increases the value of the evaluation expression is used.

  FIG. 2 is a principle configuration diagram of the present invention.

The present invention (Claim 1 ) is an experience information evaluation apparatus using a keyword of interest (hereinafter referred to as a keyword of interest),
When using a phrase that expresses the use or experience / experience of things as an experience expression,
When the attention keyword is input, the experience expression dictionary storage means 21 in which the relationship between the attention keyword and the experience expression is stored is selected, the experience expression related to the attention keyword is selected, and the experience expression storage means 20 is selected. Storing experience expression selection processing means 300;
Keyword distance that measures the appearance position relationship between the keyword of interest and the experience expression selected by the experience expression selection processing means 300 and stored in the experience expression storage means 20 in each text document of the input text document group. Measuring means 310;
Each text document of the input text document group is evaluated based on the relationship between the attention keyword and the experience expression and the appearance position relationship with the experience expression measured by the keyword distance measuring unit 310, and the evaluation result and the input text document are associated with each other. , an evaluation unit 320 for outputting the document storage means 22 Te have a,
Evaluation means 320
Based on the strength of the relationship between the keyword of interest and the experiential expression, the ranking calculation means for determining the order relationship between the text documents in the input text document group,
Ranking calculation means
As a ranking evaluation formula,
The value changes depending on the location of the attention keyword and experience expression,
In the input text document, when the experiential expression related to the keyword of interest is described before the keyword of interest, the value of the evaluation formula becomes low,
In the input text document, when an experiential expression related to the keyword of interest is described after the keyword of interest, an evaluation expression that increases the value of the evaluation expression is used.

Further, the present invention (Claim 2 ) is a ranking calculation means of the experience information evaluation apparatus according to Claim 1 ,
As a ranking evaluation formula,
The closer the distance measured from the keyword of interest and the appearance location of the experience expression, the higher the value of the evaluation formula,
An evaluation formula is used in which the value of the evaluation formula increases as the weight obtained by the appearance frequency of the experience expression related to the keyword of interest increases.

Further, the present invention (Claim 3 ) is a ranking calculation means of the experience information evaluation apparatus according to Claim 1 ,
When the update date of the input text document is available, based on the freshness of the update date, the value of the evaluation formula is increased as the input text document is newer, and the order between the text documents in the input text document group is increased. Determine the relationship.

Further, the present invention (Claim 4 ) is a ranking calculation unit of the experience information evaluation apparatus according to Claim 1 or 3 ,
When the weight of the experience expression is stored in the experience expression dictionary storage means, the weight is applied to the ranking evaluation expression, and the value of the evaluation expression is increased as the weight is higher.

The present invention (Claim 5 ) is an experience information evaluation program for causing a computer to function as each means constituting the experience information evaluation apparatus according to any one of Claims 1 to 4 .

The present invention (Claim 6 ) is a computer-readable recording medium storing the experience information evaluation program according to Claim 5 .

  As described above, in the present invention, it is possible to prioritize a set of documents related to a keyword using an experience expression corresponding to the keyword that the user pays attention to. It is possible to rank so that documents containing information are ranked higher.

  Further, when the update date of the input text document is available, in addition to the above, it is possible to rank so that a fresher document comes to the top.

According to the present invention, in the information as in Example 1) of the above-mentioned “Problem to be Solved by the Invention”,
Experience related to "restaurant": I went, I've been, I tried it, I booked it,
Experiential expressions related to "mobile phone": bought, tried, tried,
Based on the relationship between keywords and experience expressions like
Although there is an experience expression related to “Restaurant”, there is no experience expression related to “Mobile phone”, so Example 1) is given priority as experience information related to “Restaurant” Although the ranking is higher, it is determined that the experience information is not related to the “mobile phone”, and the priority ranking is lowered.

  In this way, the priority order can be changed even for the same document depending on the relationship between the keyword of interest and the experience expression.

  This makes it easier to obtain experience information related to the keyword of interest than in the prior art, and is particularly useful when the experience information of people such as reputation information has value.

  Hereinafter, embodiments of the present invention will be described with reference to the drawings.

  FIG. 3 is a diagram for explaining the outline of the present invention, and FIG. 4 shows a configuration of an experience information evaluation apparatus using a keyword of interest in one embodiment of the present invention.

In the present embodiment, as shown in FIG. 3, when a text document group in which one or more text documents are gathered and one or a plurality of attention keywords are input,
(1) Select experience expressions related to the keyword of interest,
(2) Calculate the strength of the relationship with the experience information related to the keyword of interest by measuring the distance,
(3) Ranking is performed.

  Hereinafter, this embodiment will be described in detail.

  The apparatus shown in FIG. 4 includes a computer 30 and an experience expression storage unit 20, an experience expression dictionary storage unit 21, and a document storage unit 22 that are connected to the computer 30 via a network 40 or are built in the computer 30. It is configured.

  The computer 30 includes a memory including a RAM, a ROM, and a magnetic disk, a CPU, a display unit 11 using a display, and an instruction input unit 12 including a mouse and a keyboard, and is realized by a software program executed by the CPU. An experience expression selection processing unit 31, a keyword distance measurement processing unit 32, and a ranking calculation processing unit 33 are provided.

  FIG. 5 shows an example of the experience expression dictionary storage unit in one embodiment of the present invention.

  For example, as illustrated in FIG. 5, the experience expression dictionary storage unit 21 includes “category names” such as place names and person names, “phrases” corresponding to keywords given as keywords of interest, and “related experience expressions” for the phrases. It is configured.

  Experiential expression corresponds to a phrase that expresses experience or experience of things. For example, when the phrase “restaurant” is given as a keyword of interest, the related experience expression is “I went (5), I ’ve been (5), I tried it (5), I made a reservation ( 3), ... " Here, the number written in the parenthesis is the weight of each experience expression, and represents the strength of how deeply related the phrase is. This weight is not essential. If the keyword of interest does not exist in the dictionary, a related experience expression can be searched from the “category name” column. In addition, the phrase to be included in the “category name” may be determined using a classification dictionary or the like, or may be appropriately created when constructing a device according to the purpose of use. However, this “category name” field is not essential. All these information are given in advance.

  In the experience expression, the expression itself may be stored as shown in FIG. 5, or the original form of parts of speech {going, seeing, eating, etc.} and the corresponding expression form have been tried { Etc.} may be stored. Also, the rule of combination of parts of speech after the original part of speech {go, see, eat, etc.] and the corresponding morphological analysis as shown in Fig. 6 {Verb stem + verb inflection + verb suffix + verb stem conjunctive " In some cases, “see” + termination of the verb suffix “is”, etc.} is stored, and in other cases, a mixture of the above information is stored.

  As in the example of the experience expression storage unit 20 in FIG. 7, the experience expression storage unit 20 stores the keyword of interest and the experience expression related thereto as a set. When a weight is given to each experience expression as shown in FIG. 5, the weight is also stored together with the experience expression. In the experience expression storage unit 20, information may be stored in advance, but information may be stored or deleted as needed during the process.

  The document storage unit 22 stores the position information of the keyword of interest described in the given text document group and each of the text documents, the position information of the experience expression related to the keyword of interest, and the relationship between the text document and the keyword of interest. The rank value etc. which show are stored. Either a text document group is stored in advance, or a text document is stored at any time when a text document is given. Then, in the process of processing, the location information and experience of the keyword of interest Expression position information, rank values, etc. are stored as needed.

  The experience expression storage unit 20, the experience expression dictionary storage unit 21, and the document storage unit 22 operate to realize the present invention by executing processing described below based on the apparatus configured as described above. . In this embodiment, the hot keyword is “hot spring”, and an example of experience information ranking using the hot keyword is shown.

  The input text is, for example, as shown in FIG. The output is the rank value of each input text document. Details are shown in each processing unit.

[1] Experience expression selection processing unit 31:
Based on the attention keyword input from the instruction input unit 12, a process of selecting an experience expression related to the attention keyword is performed. There are various narrowing methods, but in this embodiment, description will be made with reference to FIG. FIG. 9 is a flowchart of experience expression selection processing according to an embodiment of the present invention. It is assumed that S (S> 0) attention keywords are input to the experience expression selection processing unit 31.

  Step 101) The experience expression selection processing unit 31 substitutes 1 for a variable k when a keyword of interest is given as an input.

  Step 102) Refer to the “phrase” field and the “related experience expression (weight)” field of the experience expression dictionary storage unit 21 as shown in FIG. 5 to find the “word / phrase” field in which the k-th keyword of interest is stored. Then, the experience expression in the “Related experience expression (weight)” column accompanying it is extracted.

  Step 103) The related experience expression obtained in Step 102 is stored in the experience expression storage unit 20 together with the kth keyword of interest.

  Step 104) If the total number S of interest keywords S> k, substitute k + 1 for k and return to Step 102 (this process is repeated until S = k). If the total number of keywords of interest S = k, the process ends.

  When there is an experience expression dictionary storage unit 21 as shown in FIG. 5, if the keyword of interest is “hot spring”, the related experience expression is “I went (5), I went (5), I tried to stop by” (2), ... "can be selected. In the present embodiment, an example is shown in which each experience expression is weighted. However, the weight may or may not be present. These are stored in the experience expression storage unit 20 as shown in FIG. 7, and the process ends.

[2] Keyword distance measurement processing unit 32:
In each text document of the input text document group, the appearance position relationship between the keyword of interest and the experience expression selected by the experience expression selection processing unit 31 is measured. Although there are various methods for measuring the appearance position relationship, this embodiment will be described with reference to FIG.

  FIG. 10 is a flowchart of the keyword distance measurement process according to the embodiment of the present invention. In the following, the keyword distance measurement processing unit 32 is inputted with the keyword of interest, the number of keywords of interest (S), the experience expression related to each of the keywords of interest, one or more text documents, and the number of text documents (N). And

  Step 201) Substitute 1 for the variable a.

  Step 202) Substitute 1 for the variable k.

  Step 203) Morphological analysis is performed on the a-th input text document.

  Step 204) It is checked whether or not there is a k-th attention keyword from the output results obtained by the morphological analysis in Step 203 and an experience expression related to the attention keyword. If both the keyword of interest and the experience expression exist, the process proceeds to step 205, and if either one does not exist or both do not exist, the process proceeds to step 207.

  Step 205) If the kth keyword of interest and the experiential expression associated therewith are found in the input text document in Step 204, the distance is measured from each appearance position. The distance here refers to the distance between the k-th attention keyword and the experience expression related thereto. When there are a plurality of experience expressions related to the kth keyword of interest, the distance between each experience expression and the kth keyword of interest is measured. Similarly, when there are a plurality of k-th keywords of interest, the distance between each keyword of interest and the related experience expression is measured. In this way, even if the same attention keyword or the same experience expression is used, if the appearance position is different, the distance is measured as a different one.

  There are various measurement methods. For example, there is a method of measuring the number of bytes by counting the position where the attention keyword appears and the position where the experience expression appears as the number of bytes. There are methods such as counting by the number of characters, sentences, and morpheme elements instead of bytes.

  Step 206) Store each experience expression and distance related to the kth keyword of interest in the document storage unit 22. Here, if no text document is stored in the document storage unit 22, the a-th text document is also stored.

  Step 207) If the total number S of interest keywords S> k, the process proceeds to Step 208, otherwise the process proceeds to Step 209.

  Step 208) Substitute k + 1 for k and go to Step 204.

  Step 209) If the total number of input texts N> a, go to Step 210 and repeat this process until N = a. When N = a, the process ends.

  Step 210) Substitute a + 1 for a and go to Step 202.

  In the present embodiment, an experience expression related to each of the keywords of interest is given as an input, but the experience expression storage section 21 is included in the method of extracting from the experience expression storage section 21 at the beginning of the process or in the process of step 204. It is also possible to take out from

  In the present embodiment, a text document is given as an input. However, when a text document is stored in the document storage unit 22 in advance, a method of extracting a text document from the document storage unit 22 at the beginning of the process, In the process of step 203, a method of taking out a necessary text document from the document storage unit 22 may be used.

  In the present embodiment, morphological analysis is performed in the process of step 203, but this process is not essential. However, when the experience expression subjected to the morphological analysis is stored in the experience expression dictionary storage unit 21 or the experience expression storage unit 20, the process of step 203 is necessary. When the morphological analysis is not performed in step 203, the process can be realized by skipping the process of step 203 and searching for the keyword of interest and the experiential expression by the character string matching from the input text document in the process of step 204.

  In this embodiment, the distance between the experience expression and the attention keyword is measured in step 205, but only the position of the attention keyword and the experience expression is confirmed without measuring the distance in this step. There is also a method of storing the position information in the document storage unit 22. The location information of the keyword of interest and experience expression at this time is the number of characters, the number of bytes, the number of elements of morphological analysis, etc. It corresponds to.

  As an input text document, a document like the text document input example shown in FIG. 8 is given, and “Experienced (5), Goed (5), I tried to stop as an experiential expression related to the keyword“ hot spring ”. The time when (2),... When the morphological analysis of the input text document of FIG. 8 is performed, it is as shown in FIG. Since the morphological analysis engines are various, the analysis results may be different. As can be seen from FIG. 6, there is a hot keyword “hot spring” and an experience expression “I went”. Here, in the input text document, it is an expression of experience “I went”, but “I tried” is a polite word of “I tried”, so it is considered the same. Since there are various notations in Japanese, it is appropriate to predetermine that honorifics including polite words are the same as expressions that do not use honorifics as rules for experiential expression.

  As an experiential expression, the rule of combination of the part of speech {go, see, eat, etc.} and the corresponding part of speech after morphological analysis {verb stem + verb inflection + verb suffix + verb stem + "view" + If the verb suffix “sata” etc. is given as an input}, the combination of the part of speech obtained from the morphological analysis result of the text document and the verb prototype (verb stem) or adjective ( Examine the adjective stems to see if there is an experiential expression.

  Next, the distance is measured from the appearance position of the keyword of interest and the experience expression. As shown in FIG. 6, there is a single character “He” between “Onsen” and “I went”. The distance is 2 bytes if the number of bytes, 1 if the number of characters, and 1 case particle if the number of morpheme elements. If there is an experiential expression before the keyword of interest, the distance is measured with a minus sign such as -X byte, -X character, or -X element. This distance is stored in the document storage unit 22 together with the text document.

[3] Ranking calculation processing unit 33:
Based on the distance measured by the keyword distance measurement processing unit 32, the relationship between the keyword of interest and the experience expression is strong, and if the update date of the input text document is available, the input text document group Determine the order relationship. Although there are various ranking calculation methods, this embodiment will be described with reference to FIG.

  FIG. 11 is a flowchart of the ranking calculation process according to the embodiment of the present invention.

  In the following, it is assumed that the number of text documents (N) and the distance between the S keyword of interest in each text document and the associated experience expression are input to the keyword distance measurement processing unit 32.

  Step 301) Assign 1 to the variable a.

  Step 302) The rank value is calculated based on the ranking evaluation formula for the a-th input text document.

  Step 303) The rank value of the a-th input text document is stored in the document storage unit 22 in association with the text document.

  Step 304) If the number of input text documents N> a, the process proceeds to Step 305, where the process is repeated until N = a is satisfied, and if N = a, the process is terminated.

  Step 305) Substitute a + 1 for a, and go to Step 302.

  The ranking calculation processing unit 33 shows an example in which the distance between the keyword of interest and the experience expression related thereto is given as an input. In step 305, the corresponding distance from the document storage unit 22 is set as necessary. You may refer to it and use it in the ranking evaluation formula.

  When the keyword distance measurement processing unit 32 does not measure the distance between the experience expression and the keyword of interest and stores only the position information in the document storage unit 22, the following is performed.

  The distance is obtained from the relevant experience expression in the document storage unit 22 and the position information of the keyword of interest between the steps 304 and 305 or in the step 305 and used in the ranking evaluation formula in the step 305. There are various methods for measuring the distance. For example, the distance is obtained by subtracting the position of the corresponding keyword of interest from the position of the corresponding experience expression (number of characters, number of bytes, number of morpheme analysis elements, etc.). If the relevant experience expression is at the 20th character in the input text document and the relevant attention keyword is at the 10th character, the distance is 10 at (20-10).

  The ranking evaluation formula in the above step 302 is expressed by the distance between the keyword of interest and the experience expression, the appearance frequency of the experience expression, etc., and calculates a rank value that determines the order relationship between the input text documents. The rank value obtained by the ranking evaluation formula indicates that the higher the value, the more the experience information related to the keyword of interest.

  The ranking evaluation formula can also use the freshness when the update date of the input text document can be acquired, and the weight of the experience expression when the experience expression dictionary storage unit 21 stores the weight of the experience expression.

In the ranking evaluation formula, the rank value increases as the distance between the attention keyword and the experience expression is closer, and the rank value increases as the weight obtained by the appearance frequency of the experience expression related to the attention keyword increases. Further, when the update date of the input text document is used, the rank value becomes higher as the input text document is newer. When the weight of the experience expression is used in the experience expression dictionary storage unit 21, the rank value becomes higher as the weight is higher. Becomes higher. The ranking evaluation formula is
In the total number S of attention keywords, the k-th attention keyword is set to W _k
If there are U number of experience expressions related to W _k and the number of the expression exists in the input text document, the i-th experience expression is expressed as E _i (i ≦ U),
• If the number _{of E i} is present in the input text document is a P number, j th _{E i: E ij} the distance between the _{W k D kij,}
The weight of E _i is O _i ,
・ T,
・ The total number of input text documents is N
In this case, the ranking evaluation formula is represented by f (D, O, T).

  For example, this evaluation formula is as follows.

ＴまたはＯ_ｉが得られない場合、もしくはＴまたはＯ_ｉを利用しない場合は、ＴやＯ_ｉの代わりに１を掛けるような（式１´）で表すこともできる。（式１´）は、ＴとＯ_ｉの両者を利用しない式を示しているが、どちらか一方のみを利用することも可能である。

When T or O _i cannot be obtained, or when T or O _i is not used, it can be expressed by (Equation 1 ′) such that 1 is substituted for T or O _i . (Expression 1 ′) shows an expression that does not use both T and O _i , but it is also possible to use only one of them.

入力テキスト文書の新鮮度Ｔは、例えば、入力テキスト文書が複数存在した場合で、かつ、それらが更新日付（Time）を持っていた時、更新日付が新しいものほどＴの値が大きくなるように設定される。例えば、UNIX（登録商標）TIMEで更新日付が与えられている時には、
T=Time （式２）
で表すこともできる。または、全ての入力テキスト文書の更新日付の中から一番古い時刻を１とし、その時刻からの時間経過分（秒数等）だけ値を加算していく方法もある。

The freshness T of the input text document is such that, for example, when there are a plurality of input text documents and they have an update date (Time), the value of T becomes larger as the update date is newer. Is set. For example, when an update date is given in UNIX (registered trademark) TIME,
T = Time (Formula 2)
It can also be expressed as Alternatively, there is a method in which the oldest time among the update dates of all input text documents is set to 1, and the value is added by the elapsed time (seconds, etc.) from that time.

As the weight O _i , the value given in the experience expression storage unit 20 may be used as it is, but an expression such as TF-IDF that determines the weight of one input text from the entire input text document group. The value calculated using can also be used.

O _i = (“Number of appearances of experience expression E _i for target keyword W _k in text document to be analyzed” +1) × {log (N ÷ “number of documents in which E _i appears in input text document group” ) +1}
Formula (3)
Or
O _i = “the number of appearances of the experiential expression E _i for the attention keyword W _k in the text document to be analyzed” × (log “the number of documents in which E _i appears in the input text document group” +1) Equation (4)
Or
O _i = “number of appearances of experience expression E _i for keyword of interest W _k in text document to be analyzed” × “weight given by experience expression storage unit” Expression (5)
Etc.

  In the above formula (3), the weight is increased when there are few experiential expressions related to the keyword of interest in other text documents and there are many in the text document to be analyzed. In Expression (4), when experience expressions related to a certain keyword of interest are frequently used in the entire input text document, the weight of the text document having the experience expression becomes larger. In Expression (5), when many experience expressions related to the attention keyword appear in the analysis target document, the weight increases.

  In the present embodiment, an example in which a rank value is calculated based on the input text in FIG.

When calculating the ranking evaluation formula (Formula 1), there is “Experience (5)” as an experiential expression related to the hot keyword “hot spring”, and when the distance is calculated as 2 bytes,
From O = 5, D = 2, f (D, O) = 5 ÷ 2 = 2.5
It becomes.

If D _kij is negative, that is, if the relevant experience expression appears ahead of the keyword of interest in the input text, the relationship between the keyword of interest and the experience expression is considered to be light, and the corresponding experience expression Can be excluded from the calculation target by the ranking evaluation formula. Alternatively, the value may be divided by a preset value Y and set to take a lower value than when D _kij is positive. For example, the whole of the parentheses in (Expression 1) or (Expression 1 ′) is divided by Y, or D _kij itself is divided by Y.

  As described above, the calculated rank value is stored in the document storage unit 22 in association with the input text document.

  In this way, by presenting the calculated rank values of the respective input text documents in descending order, the update date is new, and related experience information can be presented higher by the keyword of interest. In addition to these, it is possible to refer to the document storage unit 22 and present the keyword of interest existing in each input text document and the experience expression related to the keyword of interest present in the input text document together with the ranking result. .

  In addition, the present invention constructs the operations of the experience expression selection processing unit 31, the keyword distance measurement processing unit 32, and the ranking calculation processing unit as programs and installs and executes them on a computer or distributes them via a network. It is possible.

  Furthermore, the constructed program can be stored in a portable storage medium such as a hard disk device or a flexible disk / CD-ROM, and can be installed in a computer or distributed.

  The present invention is not limited to the above-described embodiment, and various modifications and applications can be made within the scope of the claims.

  The present invention can be applied to a search technique for experience information using a keyword of interest, and can also be applied to a Web search classification, and is not limited to a field or a product.

  For example, it can be used to search reputation information that describes people's opinions and evaluations on the Internet, can be used as an index for consideration when purchasing or using the subject matter, or based on the date of collected reputation information. It can be applied to techniques for analyzing things that have become a recent topic.

  Moreover, in order to improve the reliability of information, it can be applied to search for people's experience information, and can be applied to product planning and marketing using the information.

It is a principle explanatory view of the present invention. It is a principle block diagram of this invention. It is a figure for demonstrating the outline | summary of this invention. It is a block diagram of the experience information evaluation apparatus using the attention keyword in one embodiment of this invention. It is an example of the experience expression dictionary memory | storage part in one embodiment of this invention. It is an example of the morphological analysis result in one embodiment of this invention. It is an example of the experience expression memory | storage part in one embodiment of this invention. It is an example of the text document input in one embodiment of this invention. It is a flowchart of the experience expression selection process in one embodiment of the present invention. It is a flowchart of the keyword distance measurement process in one embodiment of the present invention. It is a flowchart of the ranking calculation process in one embodiment of the present invention. It is an example of experience information.

Explanation of symbols

DESCRIPTION OF SYMBOLS 11 Display part 12 Instruction input part 20 Experience expression storage means, experience expression storage part 21 Experience expression dictionary storage means, experience expression dictionary storage part 22 Document storage means, document storage part 30 Computer 31 Experience expression selection processing part 32 Keyword distance measurement process Unit 33 ranking calculation processing unit 300 experience expression selection processing means,
310 Keyword distance measuring means 320 Evaluation means

Claims (6)

An experience information evaluation device that uses a keyword of interest (hereinafter referred to as a keyword of interest),
When using a phrase that expresses the use or experience / experience of things as an experience expression,
When the attention keyword is input, the experience expression dictionary storage means storing the relationship between the attention keyword and the experience expression is referred to, the experience expression related to the attention keyword is selected, and the experience expression storage means Storing experiential expression selection processing means;
In each text document of the input text document group that has been input, the appearance position relationship between the attention keyword and the experience expression selected by the experience expression selection processing means and stored in the experience expression storage means is measured. Keyword distance measuring means;
Each text document in the input text document group is evaluated based on the relationship between the keyword of interest and the experience expression and the appearance position relationship with the experience expression measured by the keyword distance measuring unit, and the evaluation result and the input Evaluation means for associating and outputting the text document to the document storage means ,
The evaluation means includes
Ranking calculating means for determining an order relationship between the text documents in the input text document group by a ranking evaluation formula based on the strength of the relationship between the attention keyword and the experience expression;
The ranking calculation means includes
As the ranking evaluation formula,
The value changes depending on the appearance position of the attention keyword and the experience expression,
In the input text document, when the experience expression related to the keyword of interest is described before the keyword of interest, the value of the evaluation formula becomes low,
In the input text document, when an experiential expression related to the attention keyword is described after the attention keyword, the evaluation expression is used so that the value of the evaluation expression is increased. Information evaluation device. The ranking calculation means includes
As the ranking evaluation formula,
The closer the distance measured from the attention keyword and the appearance position of the experience expression, the higher the value of the evaluation formula,
The target value of the evaluation expression higher weights obtained by the frequency of occurrence of the keyword to the relevant experience representation that a high experience information evaluation apparatus according to claim 1, wherein using the evaluation equation. The ranking calculation means includes
When the update date of the input text document is available, based on the freshness of the update date, the value of the evaluation formula is increased as the input text document is newer , and the text documents in the input text document group are increased . Determine the order relationship,
The experience information evaluation apparatus according to claim 1 . The rankings calculation means,
When the weight of experience expression is stored in the experience expression dictionary storage unit, the weight is applied to the ranking evaluation formula, and the higher the weight , the higher the value of the evaluation formula. The experience information evaluation apparatus according to 1 or 3 . The experience information evaluation program for functioning a computer as each means which comprises the experience information evaluation apparatus of any one of Claims 1 thru | or 4. A computer-readable recording medium storing the experience information evaluation program according to claim 5.

Images